Background: Standardized training of endobronchial ultrasound-guided transbronchial needle aspiration (EBUS- TBNA) is available only in highly developed countries. Objective: To compare the skill differences in the performance of EBUS-TBNA by pulmonologists trained under different learning methods. Methods: We performed a retrospective study of 145 EBUS-TBNA cases. We set up 3 comparative training groups: a standardized training group, a teaching group, and a self-study group. The main clinical outcomes were sensitivity, specificity and accuracy of EBUS-TBNA. Cumulative sum analysis was used to track EBUS-TBNA proficiency of the 3 groups. Results: The sensitivities of EBUS-TBNA in the different groups were 90, 88, and 67%, respectively. The accuracies of the different groups were 91, 90, and 71%, respectively. With respect to the sensitivity and accuracy, there was no significant difference between the standardized training group and teaching group. However, the self-study group was inferior to the other 2 groups. By comparing the learning curve among the groups, we found that the minimum number of operations needed to reach a 90% accuracy for the standardized training group, the teaching group and the self-study group was 17, 32 and 40 cases, respectively. Conclusion: Compared with the teaching and self-study groups, the standardized training group resulted in practitioners achieving high skill levels following the performance of a lower number of cases. Pulmonologists trained in the teaching group achieved a similar level of skills, but required more experience to reach this level. Practitioners in the self-study group did not achieve a satisfactory level of skills, despite performing a high number of cases.